THE LOCALIZATION OF Mg-Na-K-ACTIVATED

ON RED CELL GHOST MEMBRANES

VINCENT T. MARCHESI and GEORGE E. PALADE

From The Rockefeller University, New York 10021. Dr. Marchesi's present address is the National Cancer Institute, the National Institutes of Health, Bethesda, Maryland 0014

ABSTRACT The lead salt method introduced by Wachstein and Meisel (12) for the cytochemical demonstration of ATPase activity was modified and used to determine sites of activity on red cell ghost membranes. Preliminary studies showed that aldehyde fixation and standard concentrations of the capture reagent Pb(NO3)2 resulted in marked inhibition of the ATPase activity of these membranes. By lowering the concentration of Pb2+ and incubating unfixed red cell ghosts, over 50% of the total ATPase activity, which included an ouabain-sensitive, Na-K-activated component, could be demonstrated by quantitative biochemical assay. Cytochemical tests, carried out under the same conditions, gave a reaction product localized exclusively along the inner surfaces of the ghost membranes for both Mg-ATPase and Na-K-ATPase. These findings indicate that the ATPase activity of red cell ghosts results in the release of Pi on the inside of the ghost membrane at sites scattered over its inner aspect. There were no deposits of reaction product on the outer surface of the ghost membrane, hence no indication that upon ATP hydrolysis Pi is released outside the ghosts. Nor was there any clear difference in the localization of reaction product of Mg-ATPase as opposed to that of Na-K-ATPase.

INTRODUCTION

An ATPase' activated by Mg2+ , Na+ , and K+ , (4-9). Physiological studies on red cell ghosts and hereafter referred to as Na-K-ATPase, is found isolated squid axons have provided evidence that associated with cell membranes (1-3) and cell this ATPase is either part of or functionally as- fractions isolated from a wide variety of tissues sociated with the mechanism of active transport of Na+ and K+ (10, 11). Similarities between the I The following abbreviations are used in this paper: active transport mechanism and this ATPase ATPase, adenosine triphosphatase; ATP, adenosine activity include requirement for ATP and Mg 2+, triphosphate; ADP, ; CTP, + activation by Na and K+, and specific inhibition cytidine triphosphate; GTP, guanosine triphosphate, of both by the cardiac glycoside, ouabain. UTP, uridine triphosphate; Pi, inorganic ; In addition to this Na-K-ATPase, SDS, sodium dodecyl sulfate; EDTA, ethylene di- each of the aminetetraacetate; DEAE-cellulose, diethyl amino- membrane and microsome preparations men- ethyl cellulose; NADH, nicotinamide adenine di- tioned exhibits an ATPase activity which is de- 2+ , reduced; Tris, tris(hydroxymethyl) pendent only on Mg and is not inhibited by aminomethane. ouabain. This second activity, hereafter referred

385 to as Mg-ATPase, is presumably not involved in ghost membranes upon their ATPase activity was cation transport, but its exact relationship to the studied. Na-K-ATPase is not fully understood: the two activities have been ascribed to two distinct en- MATERIALS AND METHODS zymes or, conversely, considered as representing two different functional states of a single . Preparationof Red Cell Ghosts Both activities are known to be associated with Blood was obtained by cardiac puncture from the red cell membrane but their location in this adult guinea pigs under CO 2 narcosis, and ghosts membrane remains unknown. Previous investi- were prepared according to the following procedure: gators have attempted to determine sites of ATP- ase activity on membranes by using the lead-salt whole blood centrifuged 2,500 precipitation method of Wachstein and Meisel rpm, 15 min, with 0.1 MNa 3 citrate (12). However, objections to these studies have added as anticoagulant (5 ml/40 been raised on the grounds that the methods used ml whole blood) could not demonstrate Na-K-ATPase since this Plasma and Packed red cells washed activity was either destroyed by aldehyde fixation buffy coat twice in 0.9% NaCI, and re- or inhibited by the heavy metal used as capture aspirated off covered by centrifugation as reagent (13). Other objections have pointed out above; lysed in 10 vol of that, under the conditions of current cytochemical 5 mM Tris-HCI, pH 7.0, 1 mM procedures, nonenzymatic hydrolysis of ATP by EDTA,2 (a) mixed 15-20 lead could occur (14). min; centrifuged 25,000 g, 30 min; The present studies were initiated in an attempt r 1 to answer the following questions: (a) can a cyto- supernatant red cells resuspended in same chemical ATPase procedure be devised free of the discarded medium, mixed, centrifuged objections mentioned? (b) can the Na-K-ATPase 25,000 g, 10 min; of red cell membranes be demonstrated, and if so, supernatant (b) red cells suspended in where is this activity localized on the membrane? discarded 0.05 M Tris-HCl, pH 7.0, (c) can the Mg-ATPase be differentiated from the 0.50 M NaCI, 1 mM EDTA; 2 Na-K-ATPase on the basis of difference in locali- centrifuged 25,000 g, 10 min; zation on the membrane? supernatant red cells (now faint pink in We have chosen red cell ghosts as an experi- discarded color) suspended in 5 mM mental system since they consist of a single Tris-HCI, pH 7.0, 1 mM membrane and can be prepared easily, and since EDTA;2 centrifuged 25,000 g, their ATPase activity remains stable for relatively 10 min; long periods. Although compared to other tissues supernatant ghosts suspended in same me- (13) red cells are not particularly active, their discarded dium, centrifuged 25,000 g, 10 ATPase activity is well characterized and their min, membrane represents a system in which the corre- lation between active transport and ATPase ac- supernatant ghosts (now milky white) sus- tivity is reasonably secure (1, 2, 10). discarded pended in small volume (-0.5 ml/l ml pellet) of For these studies, a standard red cell ghost 0.05 M Tris-HCl, pH 7.0. preparation was made from guinea pig blood and experiments were carried out so that the most (a) This and all other operations after the first favorable conditions could be determined for the hemolysis were carried out in the cold. It was neces- cytochemical demonstration of both Mg- and sary to centrifuge the initial hemolysate at 25,000 g Na-K-. The responses of the red cell for 30 min to insure the complete sedimentation of all ghost system to prefixation by aldehydes and to the ghosts from the dense and viscous hemoglobin-con- presence of heavy metals used as capture reagents taining medium. This high speed was used in sub- sequent steps to shorten the centrifugation times. were investigated in light of the reports described (b) Treatment of ghosts with this high salt wash above. In addition, in view of data available on the

ATPases of other membrane systems (15, 16), the 2 A 0.1 M stock solution of Na 2-EDTA, neutralized effect of detergents and of the physical state of the to pH 7.0 with 2 M Tris, was used.

386 THE JOURNAL OF CELL BIOLOGY VOLUME 35, 1967 was an important step in rendering the ghosts free of assayed for Pi by a modified Fiske-Subbarow proce- most of the residual hemoglobin. dure. The assay system contained 0.2 ml acidified

The protein content of the final ghost preparations sample, 0.1 ml 2.5% NH 4 molybdate, 0.1 ml 1% was determined by the Lowry procedure (17) with Elon (Eastman Kodak Co., Rochester, N. Y.) as bovine serum albumin as the standard. reducing agent, and water to 1.0-ml volume. The optical density at 660 mg was linear with Pi concen- Ultrasonic Treatment of Ghost Membranes tration over the range used. Heat-treated ghosts (80 °, Ghost preparations were sonicated with a Branson 5 min) were used as controls so that ATP blanks Ultrasonifier for varying periods while kept immersed could be determined. All assays were carried out in duplicate. in an ice bath. Disruption of the ghosts to apparent 2 3 2 completion was monitored by phase-contrast micros- 32p RELEASE FROM [ p]-ATP: When [ p]_ copy. ATP was used as substrate, acid-washed Norit "A" (Amend Co., New York) was added to the HC104- Materials treated reaction mixture, and aliquots of supernate 32 containing nonNorit adsorbable p were dried on ATP and other were purchased from planchets and counted in a Nuclear-Chicago Corpo- Sigma Chemical Co. (St. Louis, Mo.). The disodium ration (Des Plaines, Ill.) gas flow counter. Parallel salt of ATP was dissolved in water and neutralized to chemical determinations showed that the 32p of the pH 6.8 with Tris. ATP concentrations in stock solu- supernate was in the form of Pi. tions were determined spectrophotometrically with the assumption of a molar absorbancy of 15.3 X 103 at Cytochemical Demonstration of 259 mu. All other materials used in incubation media were of analytical grade when available. A TPase Activity 3 2 P-labeled ATP was prepared enzymatically ac- The incubation media contained Tris-maleate cording to the method of Glynn and Chappell (18) buffer, pH 7.0, appropriate cations (added as and was modified as follows. Under appropriate chlorides), ATP or other substrates, and Pb(N03)2 32 conditions pi was incubated with ATP, 3-phos- at concentrations given in the legends of the figures; phoglyceric acid, 3-phosphoglyceric acid kinase, the reagents were pipetted in the above order. The glyceraldehyde 3-phosphate dehydrogenase (all pur- Pb(NOa)2 was added to the medium while the tube chased from Sigma Chemical Co.), and NADH. was shaken on a Vortex mixer to prevent precipitation. 3 2 [Y- P]-ATP was recovered from the HC10 4-treated All cytochemical media were completely clear before reaction mixture by passing the neutralized supernate the addition of the ghosts. With the exception of through a Sephadex G-10 (Pharmacia Fine Chemicals Pb(N0 3)2, the composition of the cytochemical Inc., New Market, N. J.) column (2.5 X 90 cm) and medium was identical with that of the biochemical eluting with 0.05 M Tris-HCl, pH 6.8. This procedure included optimal concentrations of ATP, 3 2 assay and provided a simple, one-step separation of [ P]-ATP 2+ , and Na+ and K+. 3 2 Mg from 32pi. [ P]-ATP was assayed by thin layer Reactions were run with unfixed ghosts or ghosts chromatography using DEAE-cellulose-coated plates prefixed in glutaraldehyde (20); in the case of unfixed and a formic acid solvent as described by Fahn et al. ghosts incubations were carried out at 300 or 370 for 3 2 (19). All of the p bound to ATP was located in the varying periods ranging from 10 to 30 min, and the r-position as determined by a chromatographic was stopped by chilling the incubation 3 2 reaction analysis of the products of incubation of [ P]-ATP mixture to 40 followed by the addition of two volumes with glucose and hexokinase. glutasaldehyde in 0.1 M cacodylate buffer, 3 2 of 4% 1-2% of the radioactivity of the [ P]-ATP stock pH 7.0. 32 was present as p i depending on the preparation and After storage in the cold for several hours (the time length of storage. could be varied without any noticeable effects), the the A TPase Assay mixture was centrifuged (40,000 g, 30 min), and pellets were washed in buffer and postfixed in 1% CHEMICAL DETERMINATION OF Pi: Assay s0O4 in either phosphate or Veronal-acetate buffer systems for ATP hydrolysis contained Tris HCI or (pH 7.0). Most of the pellets were stained in uranyl Tris-maleate buffers, appropriate cations and other acetate directly after Os04 fixation by methods agents, substrate, and usually -1 mg ghost protein described previously (21). After staining for 2 hr the in a total volume of 1.0 ml (see figure legends for pellets were dehydrated in ethanol and embedded in further details). Incubations were carried out in a Epon. Sections were cut with a Porter-Blum micro- 37 ° water bath-shaker and were terminated by tome provided with diamond knives. Most sections chilling followed by the addition of 0.2 ml of 10% were further stained with uranyl acetate and lead HC104. All materials were kept cold after this point. citrate before examination in a Siemens Elmiskop I After centrifugation, aliquots of the supernate were electron microscope operated with a 50 Axobjective

V. T. MARCHESI AND G. E. PALADE Red Cell Ghost Membranes 387 aperture and an anticontamination device. Micro- branes: they have the same dimensions and the graphs were taken at magnifications ranging from same filamentous material on their inner surfaces 10,000 to 40,000. Preliminary studies showed that (Fig. 4). Most membranes are organized in closed the staining of tissues in block with uranyl acetate or vesicles; few membrane fragments with free edges the staining of sections with combined uranyl-lead are encountered. stains did not change the appearance or location of Pb phosphate precipitates. A TPase Activity of Red Cell Ghosts In some experiments, ghosts were fixed in 0.5-2% glutaraldehyde (30 min in the cold room) so that it Since the ghosts used in this study were prepared could be determined whether prefixation affected the by a new method that employs a strong salt wash, localization of reaction products on the membranes. and since the ATPase assay system was modified At the end of fixation, the ghosts were recovered by for possible parallel use as cytochemical reaction centrifugation and washed in Tris-maleate buffer medium (these modifications included the use of a three times before incubating in the ATPase medium; Tris-maleate buffer and the presence of heavy after incubation, they were processed as described above, with double fixation with glutaraldehyde and metal salts), the following experiments were car- 0s04. ried out so that it could be determined whether the properties of our ghost ATPase, under the OBSERVATIONS conditions of our assay, were similar to the pre- viously described properties of ghost ATPase after Red Cell Ghosts conventional preparative and assaying procedures. When prepared as described, the red cell ghost ATP breakdown by the ghost preparations was suspension is milky white; by phase microscopy, found to proceed linearly with time when incuba- the ghosts appear as empty sacs which in form and tion media contained Mg2+, Mg2+-Na+-K+, or size still resemble swollen red cells. When viewed Mg2+-Na+-K+-ouabain (Fig. 5). The rates of hydrolysis in the presence of Mg 2+ and after incubation in a medium containing ATP, the ATP ghosts are shrunken and smaller in size and many Mg2+-Na+-K+-ouabain were identical, whereas the activity was greater than twofold in the assume a biconcave disc shape. + Electron micrographs confirm the impression presence of Mg2+-Na+-K without ouabain. that the ghosts are essentially empty sacs (Fig. 1). Table I shows that ATP was the substrate most The membranes of individual ghosts are entirely rapidly hydrolyzed by the ghost preparation under the incubation conditions used. The lack of reac- continuous and the usual, layered structure (unit membrane) is seen wherever the section is per- tivity of ADP suggested that only the terminal of ATP contributed to the Pi measured pendicular to the membrane plane. In addition, phosphate these membranes have filamentous material at- by chemical analysis. This was confirmed with similar incubations with y-32P-labeled tached to their inner surfaces which appears more data from prominent after incubation in a medium contain- ATP as substrate. ing ATP (Fig. 2). In some sections, the filamentous Pi added to the incubation medium did not material appears to be part of a feltlike network exert any inhibitory effect on the ATPase activity attached to the inner leaflet of the membrane. of the ghost membranes within the limits tested mM). ADP was found to compete with Although its chemical nature in unknown, this (up to 4.0 filamentous material serves as a useful marker of ATP when it was added at relatively high con- the inside surface of the ghost membranes. centrations, but the amount of ADP which was assay system When ghosts are subjected to relatively short generated by the ghost ATPase in the bursts of ultrasound (10-15 sec at the No. 6 was well below inhibitory levels. These results setting), they are broken up into small globules. indicated that the ATPase activity was not in- After longer treatment (30-60 sec) the preparation hibited by the reaction products in our assay becomes a turbid, cloudlike haze which contains system and that the addition of an ATP-generating practically no resolvable particles in the phase- system was not essential for maximal activity. contrast microscope. When such preparations are A TPase Activation by Detergents fixed and studied by electron microscopy, the hazy mass is found to consist of minute vesicles which are Detergents have been used to modify the ATPase limited by continuous unit membranes (Fig. 3) activity of different membrane preparations (15, indistinguishable from the original ghost mem- 22). Skou has reported that the Na-K-ATPase

388 THE JOURNAL OF CELL BIOLOGY VOLUME 85, 1967 FIGunRE 1 Electron micrograph of a representative area in a section through ghost membranes pelleted by high speed centrifugation (100,000 g, 30 min) and fixed in glutaraldehyde-OsOa. The ghosts appear as empty sacs bounded by continuous unit membranes. Fibrillar material is seen along the inner surfaces of the ghost membranes. X 90,000.

V. T. MARCHESI AND G. E. PALADE Red Cell Ghost Membranes 389 FIGURE 2 Small field in a red cell ghost preparation pelleted and fixed after incubation in a medium containing ATP, Mg'+, Na+, and K+ (concentrations as given in Table I). The filanlentous material attached to the inner surface of the ghost embrane appears more prominent than in nonincubated preparations. X 120,000. activity of nerve microsomes can be selectively gent, sodium dodecyl sulfate (SDS), the cationic increased by treatment with deoxycholate (15), an detergent, cetyl pryidinium chloride, and the effect that has been ascribed to the conversion of nonionic detergent, saponin. Preincubation of Mg-ATPase into Na-K-ATPase. ghosts with either SDS or saponin resulted in an So that it could be determined whether the activation of both ATPase activities at low con- ATPase activity of our ghost preparation shows centrations of detergents, while increasing the similar responses to detergents, ghosts were treated detergent concentration eventually resulted in the with varying concentrations of the anionic deter- complete inhibition of both components. Fig. 6

390 THE JOURNAL OF CELL BIOLOGY - VOIUME 35, 1967 FIGURES 3 and 4 Red cell ghosts were subjected to ultrasound for 45 sec, then centrifuged (100,000 g, 30 min), and fixed as a pellet in glutaraldehyde-s0 4. The ghost fragments still form closed vesicles, most of which are below the resolving power of the light microscope. The bar on Fig. 3 measures 0.2 p. In most cases, filamentous material is confined to the inner surfaces of the closed vesicles (Fig. 4); the arrows mark rare membrane fragments which show free edges. Fig. 8, X 30,000; Fig. 4, X 90,000. 6000 o Mg2 +(4mM) a Mg2 +_ Na+-K+(4 rnM:100 mM:20 mM) / fMg2+-Na+-K+(4 nnM:100 mM:20 mM) -5000 Ouobain (0.1 mM)

I 4000, o

- 3000 ) O 0 * °

a 2000 I E 0/ O

1000 1

2 4 6 8 10 12 14 16 Minutes FIGURE 5 [PI-ATP hydrolysis by red cell ghosts. Effects of Mg2+, Mg2+-Na+-K+ and Mg2+-Na+-K+ and ouabain. Incubation conditions: Tris-maleate (pH 7.0), 40 mM; Mg2+:Na+:K+, 4:100:20 rsM; [2P]- ATP, 3 mM; ouabain, 0.1 mM; ghost protein, 2 mg/ml, temperature 37°. Aliquots were removed from the reaction mixture at appropriate times, treated with 10% HC104 and Norit and counted as described in the Methods section. shows an activation-inhibition curve of ghost TABLE I ATPase activity caused by increasing concentra- Substrate Specificity of Ghost ATPase tions of SDS; a similar effect was seen with saponin, Nucleotide PM Pi/mg protein/hr although the degree of activation was not so marked. There was no apparent change in the ATP 1.40 structure of the ghosts after preincubation in CTP 0.17 detergents at concentrations which produced maxi- GTP 0.11 mal stimulation of ATPase activity, and all the UTP 0.08 ATPase activity remained sedimentable with the ADP 0 ghost membranes. Red cell ghosts (1 mg protein/ml) were incubated Incubation of ghosts in media containing cetyl for 30 min at 370 in Tris-maleate buffer (pH 7.0), pyridinium chloride resulted in varying degrees of 40 mM; Mg 2+ : Na+: K+; 4:100:20 mM, and nucleoside inhibition of ATPase activity at all the concentra- triphosphates, each at 3 mM. tions tested. tion of total ATPase in the 0.5 mM concentration Effect of Metal Salts on A TPase Activity range (Fig. 7). When lead was added to the Since heavy metal salts are necessary trapping incubation media in concentrations approaching agents for the cytochemical procedures, experi- (2.0 mM) those used in cytochemical procedures, ments were carried out so that their effects on only 10% of the original ATPase activity was ghost ATPases could be determined. Four metals still demonstrable. This inhibition could be re- which give of low solubility were versed in part by the addition of varying amounts tested for their effects on total ATPase activity. of EDTA (Table II). Pb2 + appeared to have a Barium and nickel salts had no effect on the ghost greater inhibitory effect on the Na-K-ATPase ATPase activity at the concentrations tested, while than on the Mg-ATPase (Fig. 8), yet the former lead and zinc showed approximately 50% inhibi- activity was still demonstrable, and remained

392 THE JOURNAL OF CELL BIOLOGY VOLUME 35, 1967 TABLE II 1.5 N 2+ C / Reversal of Pb Inhibition of ATPase by EDTA

*2 / Incubation conditions CPM release from Co 1.0 /'x [y 32P]-ATP/mg Pb2+ EDTA protein/30 min

mM mM Control ghost preparation 21,000 0.5 o 2.0 2,800 " O 0.2 3,200 0.3 0.6 0.9 1.2 1.5 0.6 4,400 SDS (mM) 1.0 8,400 2.0 15,400 FIGURE 6 Effect of sodium dodecyl sulfate (SDS) on 2+ - ghost ATPase in the presence of Mg and Mg2+-Na+ Red cell ghosts (1 mg protein/ml) were incubated K+. Ghosts (1 mg protein/ml) were incubated for 30 in Tris-maleate, 60 mM; Mg 2+:Na+:K +, 2:100:20, min at 370 in Tris-maleate (pH 7.0), 40 mM; Mg2+: 2+ mM; ATP, 2 mM; and Pb(NO3)2, 2.0 mM at 40 for Na+:K+, 4:100:20 mM (or Mg , 4 m, alone); and 5 min. Then varying amounts of EDTA, Mg 2+ ATP, 4 mM. (5 mM) and [32 p] ATP (2 mM) were added and the mixtures were incubated at 37° for 30 min. Re-

2+ 2 + lease of 32p from ATP was assayed as described in B or Ni the Methods section. 00 .5

U \ N described above supplemented with Pb2+ in vary- o 80 2 \ \ Pb ++SDS ing concentrations. In 0.1 mM Pb(NO 3 )2 , the v) 0 ·' A ghosts showed small precipitates associated with 60 ~the,. o inner surface of their membranes (Fig. 9). ok\'. Some of these precipitates were located directly EC Pb2+ or Zno2+ , against the membrane while others appeared at- o 40 Pb. or x tached to the associated filamentous material. No precipitates were found on the outer surface of

'E ghost membranes. Biochemically, both Mg-ATP- o 20 ase and Na+-K+-ATPase were fully active in this Q5 2+ a- low Pb concentration (see Fig. 8). I I I I l Precipitates of reaction product were larger and 0.2 0.4 0.6 0.8 1.0 more prominent on ghosts incubated in media Metal salt (mM) containing 0.5 mM Pb(NO3)2, a concentration at FIGURE'7 Effect of metal salts on ghosts ATPase. which 50% of the total ATPase was inhibited Ghosts ((1 mg protein/ml) were incubated for 80 min (Fig. 10). The distribution of precipitates re- at 87 inn Tris-maleate buffer (pH 7.0), 40 mM; ATP, mained the same, and the location appeared 3 mM, Ig2+:Na+:K +, 4:100:20 mM; and metal salts identical in preparations incubated in the presence 2+ + - at the concentrations indicated. NiC1 2, A; of Mg , Mg2+-Na+-K+, or Mg2+-Na+-K BaC1 2, A A; ZnC12, ---O; and Pb(N0 3) 2, ouabain. In no case were significant amounts of ---O. The effect of 0.5 mM SDS on the inhibition2 of ghost Al1[Pase fby Pt is also indicated (Pb SDSf precipitates attached to the outer surface of ghost x--X). membranes. The latter finding was somewhat surprising in the light of other studies (23-25). fully setnsitive to ouabain at Pb2 + concentrations Since it was possible that an ATPase originally in the rarange of 0.1-0.8 mM. bound to the external surface of cell membranes was lost during the relatively long and elaborate Cytoche emical Localization of preparation of the ghosts, intact red cells were A TPase Activity lysed in a hypotonic ATPase medium containing Cytoc:hemical tests for ATPase activity were 0.5 mM Pb 2+ and subsequently were incubated. In carried out by using the biochemical assay system this case, as in the previous experiments, precipi-

V. T. MARCHESI AND G. E. PALADE Red Cell Ghost Membranes 393 tates were confined to the inner surface of the ghost unfixed preparations. In such fixed ghosts, the membranes (Figs. 11 and 12). product of ATPase reaction appeared in the same 2+ location, i.e. the inner surface of the ghost mem- Incubation with Pb and SDS branes (Fig. 15), and was otherwise indistinguish- SDS at certain concentrations was found to able from that seen in unfixed ghost preparations. increase the total ATPase activity of the ghosts, as After fixation of ghosts in 2% glutaraldehyde, described earlier. In addition, SDS treatment ATPase activity was barely detectable (less than resulted in less inhibition of ATPase activity by 1-2% of control preparations). In the cytochemi- Pb2+ (Fig. 7). cal tests, such ghosts showed only a few precipitates Electron micrographs of ghosts incubated in scattered throughout their interior. SDS-Pb2+ showed the same localization of reac- Cytochemical Controls tion product on the inner surface of the membranes (Figs. 13 and 14). So that it could be determined whether non- Some cytochemical experiments were also car- specific Pb precipitates form under the conditions ried out with 32P-labeled ATP in order it could be used in our procedure, ghosts were incubated with

c I. I

2.Cal

E N- fI 0.5 a) . -- 1------Mg2+-Na+-K++ ouobain (0.1 mM) E ZL I * A & iPb2++ATP 0.2 0.4 0.6 0.8 Pb 2 + (mM) z+. FIGURE 8 Selective inhibition of ghost Na-K-ATPase by Pb Incubation conditions as described for Fig. 6. Pb2+ alone did not increase the nonenzymatic breakdown of ATP at the concentrations tested. Incubation of ghosts with Mg2+ as the sole activating cation in the presence of Pb + gave results identical with those shown in reactions containing Mg2-Na+-K and ouabain. determined whether the amount of reaction ATP and Pb2 + without any activating cations, or product bound to the ghost membranes in the with ADP in a complete Na-K-ATPase medium, presence of Pb 2+ and SDS was related to the total i.e. under conditions in which no enzymatic ATPase activity measured quantitatively. The breakdown of ATP or related substrates could be amount of Pb-32phosphate bound to the ghost demonstrated by quantitative biochemical assay. membranes was measured by counting aliquots of Precipitates were not found on the membranes in washed ghosts after incubation with [32P]-ATP in either case. the presence of 0.4 mM Pb2+ and 0.4 mM SDS. It As noted earlier, no precipitates were found was found (Table III) that Na-K-ATPase gave attached to the outer surface of the ghost mem- twice as much reaction product bound to the branes in any of the conditions investigated. This ghosts as did Mg-ATPase. Yet this increase was could be interpreted as indicating the absence of not immediately evident in the final micrographs ATPase activity on the external surface of the red of the cytochemical tests. cell membrane, but it could be argued that the lack of precipitates resulted from their having been A 'Pase Activity of Glutaraldehyde- removed from the outside surfaces during proc- Fixed Ghosts essing for electron microscopy. So that this pos- sibility could be investigated, Pb-phosphate pre- Ghosts fixed in 0.5% glutaraldehyde before cipitates were prepared and added to a suspension assay showed only 10%1of the ATPase activity of of ghosts which were subsequently processed for

394 THE JOURNAL OF CELL BIOLOGY · VOLUME 35, 1967 + FIGURE 9 Red cell ghosts were incubated in a medium containing Tris-maleate, 40 mM; MgZ+:Na+:K , 4:100:20 mM; ATP, 4 mM, ouabain, 0.1 mm and Pb (NO3) 2 0.1 mM at 300 for 30 min and processed as described under Methods. Small precipitates are scattered along the inner surfaces of the ghosts and in association with filamentous material. X 90,000.

V. T. MAICHESI AND G. E. PALAD Red Cell Ghost Membranes 395 FIGURE 10 Red cell ghosts were incubated in a medium similar to that described for Fig. 9, but in this experiment, the Pbz+ concentration was increased to 0.5 mM and the reaction was run at 370 for 30 min. Many precipitates of Pb-phosphate are seen along the inner surfaces of the ghost membranes. Only occasional precipitates are attached to the outside surfaces of the ghosts. X 60,000.

396 THE JOURNAL OF CELL BIOLOGY VOLUME 35, 1967 FIGURES 11 and 12 Intact red cells washed in 0.15 M NaCl were lysed in a hypotonic medium containing Tris-HCI buffer 2 mM (pH 7.2); ATP, 0.8 mM; MgZ+, 0.8 mM; and Pb(NO3)2, 0.5 m, and incubated at 37° for 15 min. Precipitates were localized only to the inner aspect of the ghost membrane and to the filamentous material attached to it. Fig. 11, X 60,000; Fig. 12, X 120,000. FIGURES 13 and 14 Ghosts were incubated in the media described for Table III. Precipitates were lo- 2+ calized to the inner surfaces of the ghost membranes regardless of whether the medium contained Mg or Mg2+-Na+-K+. The appearances of the ghost membranes and the location of Pb-phosphate pre- cipitates were identical with those found following incubation of ghosts in similar reaction media but with- out SDS. Fig. 13, X 60,000 (section unstained); Fig. 14, X 90,000. electron microscopy in the same way as cyto- ATPase, but the inhibition is concentration de- chemically reacted specimens. It was found that pendent and can be reduced to a tolerable level if 2 such exogenously added precipitates stuck to the the concentration of Pb + in the incubation me- ghosts throughout the preparation procedure and dium is brought down to 0.1-0.6 mM from the appeared attached to the outer surfaces (Fig. 16). 2.0-3.6 mM currently used in cytochemical tests. In this range, biochemical assays indicate that DISCUSSION Mg-ATPase is not affected and that a substantial fraction of Na-K-ATPase (90% at 0.1 mM General Approach Pb2+, and 50% at 0.4 mM Pb2+ ) is still available In these experiments, we have tried to carry out for the cytochemical reaction. Moreover, when cytochemical reactions under conditions similar ghosts were incubated in [y-32 P]-ATP and Pb2+, to those currently used in biochemical assays for the amount of radioactivity bound to ghosts as Mg-ATPase and Na-K-ATPase activities in vari- reaction product was found to increase by -100% ous membrane preparations. This approach gave upon Na + K+ stimulation (Na-K-ATPase) us the opportunity of determining systematically over that obtained in the presence of Mg 2+ only what type of activity, and how much of it, were (Mg-ATPase). Yet, in the electron micrographs of

TABLE III 2 Effect of Pb2 + on the Amount of 32 p Bound to Ghost Membranes Following Incubation with [y-3 p]-A TP in the Presence of SDS

Incubation conditions cmrop y-32 P/mg protein/30min

2 + Intact ghosts Mg 2+ Pb 1,000 2 + + 2 + Intact ghosts Mg + - Na - K Pb 2,400 2 + Intact ghosts No cations Pb 210 + Intact ghosts Mg 2+ - Na+ - K No Pb2+ 530 2+ Heated ghosts (80 °, 5 min) Mg 2+ - Na+ - K+ Pb 400

Red cell ghosts (1 mg protein/ml) were incubated in a medium containing Tris-maleate buffer (pH 7.0), + 2 40mM;eitherMg2+,4mM or Mg2+:Na+:K , 4:100: 20mM; [3 P]-ATP, 3 mM;SDS,0.4mM; and Pb (NO3)2, 0.4 mM (except when omitted). After incubation at 370 for 30 min, the preparations were chilled and centrifuged. The recovered ghosts were washed twice in Tris maleate buffer, 50 mu. Membrane asso- ciated radioactivity was determined, as described in the Methods section.

available for cytochemical reaction in each experi- these preparations, no great and consistent differ- mental situation. ence in the amount of Pb-phosphate precipitates Most experiments were carried out on fresh, was clearly seen. The increase is apparently too unfixed red cell ghosts, since we ascertained that small to be easily detectable under unfavorable the incubation of such ghosts did not result in conditions caused by uncontrolled variations in the detectable structure damage and since we found thickness of the sections and by some variation in that fixation by glutaraldehyde caused marked the size and distribution of reaction product (>90%) ATPase inhibition. deposits. The procedure we have used for the cyto- A TPase Localization chemical demonstration of ATPase activity is a modification of the method proposed by Wachf Regardless of the type of activity tested, i.e. stein and Meisel (12), and involves the addition o- Mg-ATPase, Na-K-ATPase, or Na-K-ATPase Pb ions to the incubation medium to trap the Pi in the presence of ouabain, the deposits of reaction liberated upon ATP hydrolysis, as electron-opaque product were always found on the inner side of the Pb-phosphate, at or near the site of its release. ghost membrane, in immediate or close contact with the latter's inner dense leaflet. Deposits also A TPases Demonstrated appeared attached to, or in contact with, the fine Pb ions were found to inhibit the ATPase filaments that form a feltwork on the inner aspect activity of the ghosts, especially their Na-K- of the membrane, and in this location they oc-

V. T. MARCHESI AND G. E. PALADE Red Cell Ghost Membranes 399 FIGURE 15 Red cell ghosts were fixed in 0.5% glutaraldehyde before incubation in a medium containing ° Tris-maleate, 40 mM (pH 7.0) Mg2+:Na+:K+, 4:100:20 mM; and Pb(N0 3)2, 0.5 mM at 37 for 30 min. Precipitates were confined to the inner surface of the ghosts and were identical with those seen in un- fixed ghosts. Arrows mark myelin figures. X 60,000. curred at distances up to 300 A from the inner In addition the sites marked by reaction product leaflet of the membrane. Since precipitates ap- deposits appear to be more numerous than esti- peared at random over the inner surface of the mated from physiological data (26). ghost membrane, there was no indication that the Experiments carried out with red cells lysed in ATPase activity is localized to a few selective sites. the incubation medium and with Pb-phosphate

400 THE JOURNAL OF CELL BIOLOGY VOLUME 35, 1967 added to the medium indicate that the absence of from the outer leaflet of the membrane. Hence, it reaction products on the outer surface of the ghost can be concluded that our procedure identifies the membranes is not the result of losing either enzyme compartment in which Pi is released upon en- (during ghost preparation) or reaction product zymatic splitting of ATP and that this compart- (during ghost processing for electron microscopy) ment is the same for ATPase activity under all

FGURE 16 Red cell ghosts were incubated in a medium to which preformed Pb-phosphate precipitates were added in order that it could be determined whether such precipitates would stick to the ghost surfaces and remain adherent throughout processing for electron microscopy. Precipitates can be seen bound to the outside surface of the ghosts. The membrane appears to dimple in at the sites of precipitate adsorption (arrows). Such dimples were not seen in ghost incubated in ATPase media. X 120,000.

V. T. MARCHESI AND G. E. PALADE Red Cell Ghost Membranes 401 conditions tested. The enzyme appears to be part membrane. Since these observations were made on of the cell membrane and our evidence suggests tissues fixed in aldehydes before cytochemical that the enzyme site active in ATP hydrolysis is incubation, and since the latter was carried located in the inner dense leaflet of the plasma- through in the standard Wachstein-Meisel medium lemma. which contains 3.6 mM Pb(NO3)2, it is likely that The presence of reaction product deposits on the the absence of reaction product on the inner subjacent filamentous feltwork is more difficult to surface was the result of inhibition of ATPase interpret: it may be caused by inefficient trapping activity by fixation, high Pb2+ concentration, or of Pi , or it may indicate the existence of ATPase both. As far as precipitates on the external surfaces in the filaments. Preliminary observations on this of red cells are concerned, they were usually filamentous material have been reported elsewhere noticed on red cells in situ, i.e. within the blood (27) and studies thereon are continued. vessels of different tissues. Since an ATPase as- sociated with the cell membranes of the endo- Bearing on Previous Physiological Data thelium is not completely inhibited by aldehyde To the extent to which they can be ascribed to fixation (23-25), it is possible that reaction product Na-K ATPase, our results are in agreement with generated by this ATPase was adsorbed non- physiological data which indicate that only intra- specifically to the red cells' outer surfaces. As cellular ATP is used in cation transport. They also described earlier, Pb-phosphate precipitates will support transport mechanisms which postulate the adhere firmly to the outer surfaces of red cell

intracellular release of the terminal P0 4 of ATP, ghosts and may react in the same way with the regardless of other features involved such as a membranes of intact red cells. Alternatively, the multistep reaction or a two step reaction in which reaction may be caused by phosphatases adsorbed a Mg2+-activated phosphorylation reaction is from the plasma on the outer surfaces of the red followed by a K+-activated phosphatase step (28). cells. Yet, these considerations do not exclude the There is no agreement, however, between our possibility that red cells of some species have findings and the view that Mg-ATPase (considered ATPase activity on their external surface: in a in this interpretation as nonspecific apyrase) is recent study by Tooze (31), reaction products were located on the outer surface of the membrane localized to the outer surface of isolated amphibian (10). red cells, while the outer surface of rat and rabbit Since the location of reaction product is similar red cells were nonreactive. for Mg-ATPase and Na-K-ATPase, our observa- By lowering Pb2+ concentration we have prob- tions provide no further insight into the relation- ably reduced the efficiency with which released Pi 2+ ship of these two activities. Recent studies suggest is captured by Pb . This could result in Pi that Mg-ATPase can be converted into Na-K- diffusion away from the site of its production before ATPase by a variety of treatments (2, 15, 22, 29, its capture and precipitation. Fortunately, in our 30), and such findings have been taken to indicate experiments Pb2 + concentration in the media was that the two activities represent different func- apparently sufficient to precipitate the released Pi tional states of a single enzyme. Yet, since neither reasonably close to the sites of its production on the ATPase has been isolated and characterized in ghost membrane. We base this assumption on the terms of its molecular properties, there is no firm following points. evidence to establish whether one or two (a) The localization of the Pb-phosphate pre- are present in these membranes. cipitates was roughly the same regardless of the concentration of Pb2+ in the incubation medium Bearing on Previous (over the range 0.1-0.8 mM). Cytochemical Observations (b) The amount of Pb- 3 2 phosphate bound to the Previous attempts to study ATPase localization ghost membranes as ATPase reaction product was in mammalian red blood cells by lead-salt pre- found to be proportional to the total ATPase cipitation methods have produced results which do activity of the reaction. not agree with our findings. In these studies, (c) Reaction products were found only on one precipitates were generally seen attached to the side of the ghost membrane. Extensive Pi diffusion outside surfaces of red cells, and no reaction from the sites of its production should give, if product was detected on the inside surface of the present, more precipitates on the outer surface of

402 THE JOURNAL OF CELL BIOLOGY VOLUME 35, 1967 the membranes since the ghosts are known to be such low Pb 2+ concentrations; this means that the permeable to Pi (32). conditions used in this study might not be suitable The extent of Pi diffusion away from its site of for other experimental systems. production probably depends on the relative activity of the enzymes involved. The specific The excellent technical assistance of Miss Christine ATPase activity of our ghost preparation, al- Schillig is gratefully acknowledged. though similar to corresponding figures available This work was supported in part by the United in the literature, is rather low: 1.4 M P/mg States Public Health Service grant No. HE 05648-06. protein/hr. It is likely that this low activity has Received for publication 20 March 1967; revision accepted been an important factor in our successful use of 28 July 1967.

REFERENCES

1. POST, R. L., C. R. MERRITT, C. R. KINSOLVING, 12. WACHSTEIN, M., and E. MEISEL. 1957. Histo- and C. D. ALBRIGHT. 1960. Membrane adeno- chemistry of hepatic phosphatases at a physi- sine triphosphatase as a participant in the ological pH. Am. J. Clin. Pathol. 27:13. active transport of sodium and potassium in 13. BONTING, S. L., L. L. CARAVAGGIO, and N. M. the human erythrocyte. J. Biol. Chem. 235: HAWKINS. 1962. Studies on sodium-potassium- 1796. activated adenosinetriphosphatase. IV. Cor- 2. DUNHAM, E. T., and I. M. GLYNN. 1961. Adeno- relation with cation transport sensitive to sine triphosphatase activity and the active cardiac glycosides. Arch. Biochem. Biophys. 98: movements of alkali metal ions. J. Physiol. 413. 156:274. 14. ROSENTHAL, A. S., H. L. MOSES, D. L. BEAVER, 3. SEN, A. K., and R. L. PosT. 1964. Stoichiometry and S. S. SCHUFFMAN. 1966. Lead ion and and localization of adenosine triphosphatase phosphatase histochemistry. I. Non-enzymatic dependent sodium and potassium transport in hydrolysis of nucleoside phosphates by lead the erythrccyte. J. Biol. Chem. 239:345. ion. J. Histochem. C(tochem. 14:698. 4. SKOU, J. C. 1957. The influence of some cations 15. SKou, J. C. 1962. Preparation from mammalian on an adenosine triphosphatase from peripheral brain and kidney of the enzyme system in- nerves. Biochim. Biophys. Acta. 23:394. volved in active transport of Na+ and K+. 5. KINSOLVING, C. R., R. L. POST, and D. L. Biochim. Biophys. Acta. 58:314. BEAVER. 1963. Sodium plus potassium trans- 16. ASKARI, A., and J. C. FRATANTONI. 1964. Effect port adenosine triphosphatase activity in kid- of monovalent cations on the adenosinetri- ney. J. Cellular Comp. Physiol. 62:85. phosphatase of sonicated erythrocyte mem- 6. LANDON, E. J., and J. L. NORRIS. 1963. Sodium- brane. Biochim. Biophy-. Acta. 92:132. and potassium-dependent adenosine triphes- 17. LOWRY, O. H., N. J. ROSEBROUGH, A. L. FARR, phatase activity in a rat-kidney endoplasmic and R. J. RANDALL. 1951. Protein measure- reticulum fraction. Biochim. Biophys. Acta. 71: ment with the Folin phenol reagent. J. Biol. 266. Chem. 193:265. 7. SCHWARTZ, A. 1962. A sodium- and potassium- 18. GLYNN, I. M., and J. B. CHAPPELL. 1964. A stimulated adenosine triphosphatase from car- simple method for the preparation of 32p_ diac tissues. I. Preparation and properties. labelled of high specific Biochem. Biophys. Res. Commun. 9:301. activity. Biochem. J. 90:147. 8. SCHWARTZ, A. 1963. A Na+ K+-stimulated 19. FAHN, S., R. W. ALBERS, and G. J. KOVAL. 1965. adenosine triphosphatase in "microsomal" Thin-layer chromatography for the separation fractions from rat liver. Biochim. Biophys. Acta. 67:329. of nucleotides. Analytical Biochem. 10:468. 9. ERNSTER, L., and L. C. JONES. 1962. A study of 20. SABATINI, D. D., K. BENSCH, and R. J. BARRNETT. the nucleoside tri- and diphosphate activities 1963. Cytochemistry and electron microscopy. of rat liver microsomes. J. Cell Biol. 15:563. The preservation of cellular ultrastructure and 10. HOFFMAN, J. F. 1962. Cation transport and struc- enzymatic activity by aldehyde fixation. J. ture of the red cell plasma membrane. Circula- Cell Biol. 17:19. tion. 26:1201. 21. FARQUHAR, M. G., and G. E. PALADE. 1965. Cell 11. SKOu, J. C. 1965. Enzymatic basis for active junctions in amphibian skin. J. Cell Biol. 26:263. + + transport of Na and K+ across cell mem- 22. JARNEFELT, J. 1964. Conversion of the Na and branes. Physiol. Rev. 45:596. K+ independent part of the brain microsomal

V. T. MARCHESI AND G. E. PALADE Red Cell Ghost Membranes 403 ATPase to a form requiring added Na+ and 28. ALBERS, R. W., S. FAHN, and G. J. KOVAL. 1963. K+. Biochem. Biophys. Res. Commun. 17:330. The role of sodium ions in the activation o 23. NOVIKOFF, A. B. 1962. Symposium on the plasma Electrophorus electric organ adenosine triphos- membrane. Circulation. 26 (Suppl.):1126. phatase. Proc. Natl. Acad. Sci. U. S. 50:474. 24. MARCHES], V. T., and R. J. BARRNETT. 1963. 29. HOKIN, L. E., and D. REASA. 1964. Effect of The demonstration of enzymic activity in preincubation of erythrocyte ghosts on oua- pinocytotic vesicles of blood capillaries with bain-sensitive and ouabain-insensitive adeno- the electron microscope. J. Cell Biol. 17:547. sine triphosphatase. Biochim. Biophys. Acta. 25. MARCHESI, V. T., and R. J. BARRNETT. 1964. 90:176. The localization of nucleoside phosphatase 30. SKOU, J. C., and C. HILBERG. 1965. The effect activity in different types of small blood vessels. of sulphydryl-blocking reagents and of urea on J. Ultrastruct. Res. 10:103. the (Na+ + K+)-activated enzyme system. 26. TOSTESON, D. C., P. COOK, and R. BLOUNT. 1965. Biochim. Biophys. Acta. 110:359. Separation of adenosine triphosphatase of HK 31. TOOZE, J. 1965. An investigation by electron and LK sheep red cell membranes by density microscopy of the nucleoside phosphatase ac- gradient centrifugation. J. Gen. Physiol. 48: tivity of amphibian and mammalian erythro- 1125. cytes. J. Cell Biol. 26:209. 27. MARCHESI, V. T., and G. E. PALADE. 1966. 32. PAssow, H. 1964. Ion and water permeability of Protection by ATP of the structure and the red blood cell. In The Red Blood Cell. C. ATPase activity of red cell ghosts against Bishop and D. M. Surgenor, editors. Academic tryptic digestion. J. Cell Biol. 31:72A. Press Inc., New York. 71.

404 THE JOURNAL OF CELL BIOLOGY VOLUME 35, 1967